Don't believe what you read in the newspapers: the opinion polls did not get the election right
STEPHEN GLOVER
There is a general assumption that the opinion polls got it right. The only dissenting voices I have come across are John Curtice in the Independent and Alan Travis in the Guardian. Otherwise journalists say nothing or in some instances even commend the polls. After all, the polls predicted that the Tories would be slaughtered, and they were. The Association of Professional Opinion Polling Organisations has even issued a self-congratulatory statement claiming to have confounded its critics.
But the polls did not get it right. Or if they did it was only in the sense that someone may predict that A will knock out B in the first round of a boxing match, whereas in the event the contest goes to six rounds with B landing the odd punch on A before being counted out. For this is the extraordinary fact. Not one opinion poll among the many during the election campaign put Labour's lead as low as the nine points it achieved on the night. That's right. Not one poll.
A good number of polls put the Labour lead at over 20 points, and a few over 25. On 23 May the Sun led with a story devastating to the Tories. A Gallup poll put Labour on 55 per cent of the vote with the Tories on 25 and the Lib Dems on 14. 'That would give PM Tony Blair an astonishing 355 majority,' the newspaper informed us. 'The Conservative wipe-out would slash the number of William Hague's MPs to as few as 82.' In the event 166 Tories were returned. The Sun quoted various authorities. Professor Paul Whiteley of the British Election Survey at Essex University (which put Labour's lead at 30 per cent) said, 'The chances of all the polls being wrong are many millions to one.' Bob Worcester of MORI thought that, 'No matter how you do your sums, we are looking at a meltdown.'
This particular Gallup poll was only an extreme example of a general tendency. MORI also gave Labour consistently huge leads. Four days before election day the Sunday Telegraph declared on its front page: 'MORI poll gives Labour 23-point lead as Conservative recriminations begin.' MORI put Labour on 50 per cent, the Tories on 27 per cent and the Lib Dems on 17 per cent. (The actual figures were 42, 32.7 and 18.8 respectively.) The admirable John Curtice mentioned above has calculated that the average Labour lead given by all the opinion polls throughout the election was as follows: MORI 20 points; NOP 18 points; Gallup 16 points; ICM 15 points; and Rasmussen, used by the Independent but ignored by the BBC in its poll of polls, 12 points. To be fair, all the pollsters were somewhat closer to the actual Labour lead of nine points in their final polls. NOP and Gallup were at 17 points; MORI at 15; and ICM at 11.
I decided to telephone the eminence Rise of pollsters, MORI's Bob Worcester, to see what he had to say for himself. If I halfexpected a degree of uncertainty, a feeling even that the polls had got it slightly wrong, I was mistaken. Mr Worcester was characteristically upbeat. I shall attempt to summarise his arguments. 1) The polls were predicated on a lower turnout than in 1997, but not as low at it was. 2) Polls are only a snapshot. They do not predict. It follows that a poll conducted two weeks before election day does not tell us what will happen on election day. John Curtice's averages are therefore meaningless. 3) There seems to have been a significant swing to the Tories very late in the campaign. 4) Journalists do opinion polls a disservice by concentrating on the gap between parties rather than on the share of vote. 5) If you allow for a three-point margin of statistical error, plus or minus, the polls were not out of line. Mr Worcester did concede, however, that the Sunday Telegraph poll on 3 June which gave Labour a 23-point lead was 'at the extreme'.
There seems, as one might expect, to be an element of special pleading about all this. There probably was a swing to the Tories in the final week — a feather in William Hague's cap — but most polls still overestimated Labour's lead by a significant margin. And can we really believe that the polls which found a Labour lead of 25 per cent-plus a couple of weeks before election day were reflecting anything like the true state of affairs? I find that very difficult to swallow. As for Mr Worcester's complaint that journalists concentrate too much on the gap between parties, that is perhaps a little disingenuous. Pollsters allow their data to be portrayed in a certain way.
Throughout most of the campaign many of the polls predicted a Tory wipe-out — or meltdown to quote Mr Worcester in the Sun — which did not exactly materialise. My submission is that the political debate took place against a false backdrop. But of course you will never get the pollsters to
concede fault. They attempt to take refuge in the argument that what the polls said was true at the time, give or take the usual margin of error. I don't believe it. But I imagine we will go on listening to the pollsters whether they improve their methodology or not. Prophets go on prophesying even when their prophecies come unstuck.
Afew years ago I unveiled my theory of baldness in these pages while admitting that my colleague, the venerable Alan Watkins, may have come to a similar conclusion working away in a different laboratory. The theory is that in the television age in Britain and America it is well-nigh impossible for a bald leader to prevail. The last bald leader to be elected in this country was Winston Churchill; in the United States it was Eisenhower. By contrast, in modern France it is practically impossible to be elected president or chosen as prime minister unless you are bald. I cannot think of a single political leader during the Fifth Republic who has had a full head of hair. Mitterrand, Giscard D'Estaing, Pompidou and De Gaulle (after 1958) were all bald, as are Chirac and Fabius. People sometimes claim that Lionel Jospin is well covered, but I have seen him at close quarters and I can assure you that he isn't. It seems that in the Anglo-Saxon world baldness conveys weirdness or decrepitude in a politician, whereas in France it symbolises wisdom and, for all I know, virility.
The theory implies something so frighteningly superficial that I am sometimes inclined to disbelieve it myself. But perhaps we live in frighteningly superficial times. Elections are now determined on the television screen rather than on the stump or in the debating chamber. The image is all. After the failure of baldies Neil Kinnock and William Hague (John Smith was never put to the test) the Tories should at least consider the theory when selecting their new leader. fain Duncan Smith is spectacularly bald. On the other hand, Michael Portillo's lavish locks proliferate so much that they look as though they have to be kept in place by hair oil or the occasional perm. Perhaps Mr Duncan Smith should sit out a Portillo leadership, withdraw even further from the limelight, and ingest large quantities of a new anti-baldness wonder pill soon to be launched by GlaxoSmithKline.
Previous page
